Skip to main content

Techné AI · Free reference · Edition 3.0.0

Source Guide

A primary-source guide for this handbook, organized by topic; detailed provision-level citations and qualifications remain with each chapter.

Reviewed Download complete PDF Corrections
On this page
  1. How to verify a statement
  2. Law and regulation
  3. Standards and risk frameworks
  4. Technical evidence and practical examples
  5. Review boundaries

How to verify a statement

This is a reading map, not a claim that every source has been exhaustively reviewed or that all developments are covered. The chapter footnotes are the detailed source record: they identify the particular instrument, provision, publication or court order used for the statement.

When checking a legal proposition, record the instrument’s status, relevant version, jurisdiction, regulated role, application date and any later amendment. Publication, entry into force, general application and a transitional deadline can be different dates. A press release about an agreement is not a substitute for the adopted text.

Law and regulation

TopicStart withDetailed handbook source record
European UnionRegulation (EU) 2024/1689 and Regulation (EU) 2026/1744, followed by relevant Commission implementation materialEU AI Act
US federal policy and agency requirementsOfficial executive orders, OMB memoranda, statutes and agency publications; distinguish internal federal requirements from private-sector dutiesUS federal
US states and citiesCurrent statutory text and official regulator publications, rather than only the bill announcement or an earlier effective-date articleUS states and cities
Other jurisdictionsOfficial legislative and ministry sources; label pending bills and voluntary initiatives separatelyInternational governance
Healthcare, finance and employmentThe responsible sector regulator’s statute, rule or dated guidance, including its stated scope and statusSector-specific rules
Training data and copyrightThe actual court order, procedural posture and affected claims; do not generalize one fact-specific ruling into a universal permissionCopyright and IP
Privacy and automated decisionsApplicable data-protection law and official regulator guidance, with attention to territorial scope and exceptionsPrivacy, data and security

The September edition corrects the old handbook’s use of secondary commentary where an official source was available. It also removes earlier forecasts that no longer describe the legal position. A source link is not a substitute for an organization-specific applicability review.

Standards and risk frameworks

The handbook’s maturity-planning tables are illustrative planning aids, not an official NIST scoring system or a validated assessment instrument. A claimed alignment should identify the specific source provisions and actual evidence, not just a framework name.

Technical evidence and practical examples

The technical safety, frontier-models and privacy chapters link technical and official materials beside the relevant discussion. Evaluate the study’s task, data, system version, test conditions and limitations before applying a result elsewhere.

A vendor’s description of its own policy is evidence of that published policy, not independent proof of the model’s performance. A benchmark result is not a finding that every deployment context is acceptable.

Review boundaries

This edition is dated September 7, 2026. Sources were checked for the propositions discussed, but the handbook is not a continuously updated legal database, licensed standards collection, complete docket review or exhaustive global survey. Translations and summaries should be checked against the authoritative language where it controls.

For the changes made in this edition, see edition history. For an error or a later development, use the corrections process with the affected passage and source.

This free handbook is a dated educational reference, not a determination of your organization's obligations. Check the source, jurisdiction and role before applying a requirement. For working documents, see TalentSight Intelligence and BoardSight Intelligence.